In January 2005 the Dutch Advertisers Association (BVA) presented the results of an observational study on viewing behaviour during commercial breaks. Results from filming 100 households watching TV showed only 20% of commercial airtime is watched attentively, while 20% walk out of the room, 20% is zapped away and during the rest people stop looking at the screen and start to talk, phone or read. Similar results were obtained in another, smaller scale observational study by Mark Ritson of the London Business School, where 'the percentage of advertising watched varied from 23% to 55%' (Ritson, 2003). TV is no longer the centre of attention in the household. For effective advertising we must look for those rare moments when people pay attention. Based on an observational study by the University of Leeds, Sheila Byfield concluded 'the level of viewer involvement in programmes has an important and direct influence on whether the break will be viewed - at all' (Byfield, 2000). This paper shows how viewing duration or 'net fraction' from the people meter panel is used as a proxy for attention and how this approach was validated using a large scale study on audience attention and program appreciation.